Sense getting worse after 9 months


#1

Simply, i got this in november. For the current wattage draw its worth the cost. But the device detections is terrible. Yeah, for several months it was rock-solid identifying the fridge, freezer, and garage door. The coffe maker is confused with the stovetop. The dryer has to be in halves. Seriously? Eight months and it still can’t go 240v? The most imortant thing for me is a 5hp air compressor. It would show me if any leaks are present. I’m on the fourth -fourth!- id of this. Unknown motor, unknown motor, air conditioner, unknown, and it still can’t figure it out. This thing stargs with 19k watts and runs for two minutes. Like nothing in the house. Cant remember that?

Now in summer, it hasnt learned any of the three window a/cs. But, at least it thinks an a/c or two is the fridge. So thats good.

“Oh, this is complicated; they both draw similar loads blah blah blah”. What about sampling thousands of times a second? The fridge and a/c are identical? Really?

Oh well; i still like the sense for overall usage but detection has not improved in 9 months.


#2

@peltzr, sorry to hear that you haven’t had an ideal device detection experience :confused:. It does sound like you’re experiencing some frustrating issues. We are working on some improvements to our fridge detector that should hopefully help things there. With regards to your dryer being in halves, our new device merging feature should take care of that. Have you had a chance to try that out yet? My other suggestion would be if you feel that a device is particularly problematic, you can try deleting it. Deleting a malfunctioning device can sometimes lead to an improved re-detection of it.

Really appreciate your patience as we continue to work to improve device detection. I know it can be frustrating at times, but we’ve made a lot of progress since November (even if it doesn’t always feel like it)!


#3

I’ve had my unit for a year now and had similar problems happening (coffee maker <-> stove, single device reported multiple times, not finding 220V pool pump which is on a fixed timer, etc). I did a full factory reset in May. Yes I lost history but with all of the issues it wasn’t useful. It’s been much better since then. It took about 2 months but I’m back to where I was in terms of devices (though not all the same ones) and it finally found the pool pump (for me pool season jun-sep).

I did the full on reset as other posts indicated this was the best (only?) way to get the issues to go away.
YMMV


#4

Aha, i do now see the merge feature. I will now do a full reset and start fresh. Thanks for the tips!


#5

Update - i did the full reset and started from scratch. Lo and behold, it is far better. It got some things in a month that it hadnt gotten in nine months prior, like the toaster oven, stovetop, and lg tv. That was cool, btw. It came up with “Lg Tv” on its own. I guess thats part of network id? Also, the compressor is acurately monitored.

Summary, full reset is totally worth it. Maybe by dynamically deleting individual items and letting it relearn i wont have to full reset again.

Thanks!


#6

Yes! We’ve started to roll out TV models (thanks to network device identification data). Excited to hear that your TV was detected!


#7

Tonight I turned on my Samsung TV and Sense alerted me that it found a Samsung TV. Very cool!


#8

Have had the device installed since April 2017. Device detection has been extremely poor. have 2 AC units… the AC (Condenser) is detected but air handler is not… My Electric car is not, it was detected once as a 100W device when charging… Oven / Lights / garage door / TVs / Fridge are a far cry… i have reset the device 3 times already… just did another one a moment back, but honestly not hopeful that i will see any difference.

Incase the internet does down, i have to go and reboot the device before it can get connected again.


#9

@thevaibhav.arora, I strongly recommend giving Sense some time to gather data on the devices in your home. It needs to see devices a number of times before it can accurately identify them. Unless you see a major issue, I would let Sense do its thing for a couple months. If you feel that Sense is still having issues detecting devices, you can get in touch with our support team who will be able to look at the data your monitor has collected and help troubleshoot.

I also want to mention that Sense isn’t great at detecting electric vehicles at the moment. We’re still working on developing detecting models in that regard, although Model S’s are detected fairly well at this point.


#10

I’m getting really disappointed in Sense. I’ve had mine for 5 months and it still hasn’t detected devices that are used every day. I turn on heat lamps in my bathroom every morning for 30 minutes. It hasn’t detected them. Then I turn on a vent fan for 15 minutes. It’s a motor. Those are normally easy to detect. It still hasn’t gotten that. Why not? I haven’t heard a good explanation.

Also, many people have mentioned the need for training. There are many devices being included in the ‘always on’. I would like a process where I could turn devices off and on a series of times so Sense could learn their pattern. It would really help to reduce the number of devices in the ‘always on’ group.


#11

I normally don’t pile on and I am following the “First EV Devices!” thread closely. That said, I have detected a disproportionally high representation of Sense owners who are also EV owners. That should be motivation to make EV detection a priority especially considering that EVs are large consumers of electricity. I am highly anticipating the day my EV is discovered so I can correlate it’s usage in the context of our TOU pricing in light of the fact that there are times in the middle of the night when rates go negative.


#12

@davidferri thanks for the feedback. I’ve shared it with our data science team. Work on EV detection will pick up again soon!


#13

@BenAtSense You have carefully avoided speaking to the suggestion that people are continually doing a data delete/factory reset in hopes of better detection the next time around. I cant imagine this is what you guys are intending people to do? This is probably a question that should be addressed so that people aren’t doing it in hopes of the device working better, and being disappointed at the results. It’s openly being talked about as a work-around solution by users and not being addressed by the sense group.


#14

I tried the data delete to get some improvements in the area device detection, which had appeared to stall with half dozen devices and two being processed. That “broke” my solar Sense’s ability to correctly measure power. Three weeks passed with several full resets and numerous emails with support team.
Finally started working two weeks ago but device detection is now slow and not consistent.
So, be cautious about doing a reset, the consiquents can be trouble.!


#15

Hey @mark.bays,

If you search around, I have addressed when we recommend to a data reset, but it has been a little while so…

Generally speaking resetting your monitor should not affect device detection. After a data reset, it is possible that different devices will be detected first, but there should not be a dramatically different experience overall. Given this, we do not recommend resetting your monitor except in extreme cases (and after our support team has taken a look). If a specific device is having issues, deleting it and having it re-detected might help.

Here’s a support article we have on the topic (which mostly reiterates what I said):

Let me know if you have any other questions.


#16

As a follow-up, it’s been over a year since I’ve had Sense, and it still hasn’t detected the heat lamps or vent fan that we use every morning in the bathroom. They seem to be simple devices that should have simple electrical signatures, yet Sense still doesn’t get them. This is also early in the morning, so there aren’t a lot of other devices on to get them confused with. I would really like the people at Sense to take a look at the data and explain why these aren’t being detected.


#17

What model Samsung? Still waiting for that one.


#18

Hasn’t found either of my two Samsung TV, nor the TIVO’s, nor the DVD, nor the 6 channel stereo, in 16+ months…you have to be patient


#19

I figured I would add my .02 and mention that it seems like Sense has some great power detection hardware that can be extremely accurate in detecting power draw.

You might want to start working with your customers on a method for detecting different device types.

We all know what devices we have in our house. You could develop a learning mode where we add a device template and turn it on within a specific time period.

So far for me, Sense has not detected any of my lights (We have some room/area lights that draw over 600 watts when turned on) and only detected motors and heat sources for the most part.

I understand the problems you face when trying to isolate different devices from the noise of a busy house, especially when all of the devices are running at once, but it might help to have some base patterns to work from.


#20

Hey Randy. Unfortunately while these devices might look simple to us, Sense sees things with much greater detail that tends to erode that seeming simplicity. Have you submitted a Support ticket on this? If not, do so via support@sense.com. You can mention that you were directed there by me. They might be able to offer some help.

@andy and @markhovis73 we’ve gotten pretty good at detecting Samsung smart TVs. Are your TVs internet connected and you have NDI enabled (Settings > My Home > Data Sources)?

@arch Thanks for the feedback. Learning modes just aren’t all that feasible (we actually had one in the early days and removed it because it did not work nearly as expected). As I mentioned above, Sense sees the world quite differently than us. Simply showing it what a device is supposed to look like doesn’t work for this reason. As you note, a lot of our challenge is in isolating signal (the device you want to find) from noise (both actual power line noise as well as the devices you don’t want Sense to confuse with this one). So, training Sense with an isolated waveform wouldn’t be all that helpful as, in real-world contexts, that device is always running alongside other devices. It’d kind of be like finding a needle in a haystack, except the haystack is all needles and they are all constantly changing in shape and size. We are working hard on developing other ways to get ground truth data into Sense (similar to our inclusion of NDI), so stay tuned.